CN107450323B - Hypersonic aircraft reentry stage neural network Hybrid Learning control method - Google Patents

Hypersonic aircraft reentry stage neural network Hybrid Learning control method Download PDF

Info

Publication number
CN107450323B
CN107450323B CN201710789276.6A CN201710789276A CN107450323B CN 107450323 B CN107450323 B CN 107450323B CN 201710789276 A CN201710789276 A CN 201710789276A CN 107450323 B CN107450323 B CN 107450323B
Authority
CN
China
Prior art keywords
neural network
control
hypersonic aircraft
reentry stage
error
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201710789276.6A
Other languages
Chinese (zh)
Other versions
CN107450323A (en
Inventor
许斌
寿莹鑫
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwest University of Technology
Shenzhen Institute of Northwestern Polytechnical University
Original Assignee
Northwest University of Technology
Shenzhen Institute of Northwestern Polytechnical University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Northwest University of Technology, Shenzhen Institute of Northwestern Polytechnical University filed Critical Northwest University of Technology
Priority to CN201710789276.6A priority Critical patent/CN107450323B/en
Publication of CN107450323A publication Critical patent/CN107450323A/en
Application granted granted Critical
Publication of CN107450323B publication Critical patent/CN107450323B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention discloses a kind of hypersonic aircraft reentry stage neural network Hybrid Learning control method, the technical issues of for solving existing hypersonic aircraft reentry stage attitude control method low precision.Technical solution is the attitude mode first by hypersonic aircraft reentry stage kinetic description for Control-oriented, neural network is recycled to carry out learning scene controller to system nondeterministic function, finally use online data structure forecast modeling error, and form combination misalignment using tracking error and prediction error and carry out the update of neural network weight, promote closed loop control process neural network learning performance improvement tracking performance.Due to using neural network carry out study can lifting system adaptive ability, improve control precision；Due to carrying out neural network weight vectors complex updates, improving neural network learning accuracy in closed loop control process, improve systematic tracking accuracy using online data structure forecast error assessment neural network learning performance and in conjunction with system tracking error.

Description

Hypersonic aircraft reentry stage neural network Hybrid Learning control method
Technical field
The present invention relates to a kind of hypersonic aircraft reentry stage attitude control methods, in particular to a kind of hypersonic winged Row device reentry stage neural network Hybrid Learning control method.
Background technique
Hypersonic aircraft is due to its high-speed flight ability, so that in case of emergency realizing that " whole world reaches, the whole world is made War " is possibly realized, therefore by extensive concern both domestic and external；NASA X-43A, which makes a successful trial flight, confirms the feasible of this technology Property.During hypersonic aircraft reenters, the variation range of the angle of attack and angle of heel is big, and there are strong nonlinearities for system；Meanwhile gas Dynamic coefficient, thrust reverser controlled efficiency it is all serious depend on flight attitude, this deposits parameter in system and state all In serious Non-linear coupling.The characteristic of hypersonic aircraft reentry stage brings huge challenge to control design case, control System processed must just have strong adaptive learning ability.
Since characteristics, the aircraft manufacturing technologies such as reentry stage strong nonlinearity, fast time variant and strong uncertainty are very multiple It is miscellaneous.Common processing scheme is designed using robust adaptive.A kind of scheme is that the non-linear of system is written as linear parameterization Form, and then adaptive design is carried out, which needs to have clear cognition to obtain Parameter Expression the structure of system；Separately A kind of outer scheme is the bound that will consider nonlinear function, using the information design robust item to guarantee that system is stablized, such as 《Adaptive Dynamic Sliding Mode Control for Near Space Vehicles Under Actuator Faults " (Jing Zhao, Bin Jiang, Peng Shi, Hongtao Liu, " Circuits Systems&Signal Processing ", 2013, volume 32, the page number: 2281-2296) text had studied sliding formwork control for Near Space Flying Vehicles Device.Due to carrying out sliding mode design using the unknown upper bound information of system in design process, obtained controller has very strong Conservative, be unfavorable for high-precision hypersonic aircraft reentry stage gesture stability.
Summary of the invention
In order to overcome the shortcomings of existing hypersonic aircraft reentry stage attitude control method low precision, the present invention provides one Kind hypersonic aircraft reentry stage neural network Hybrid Learning control method.This method first reenters hypersonic aircraft Section kinetic description is the attitude mode of Control-oriented, and neural network is recycled to carry out learning scene control to system nondeterministic function Device processed, finally use online data structure forecast modeling error, and using tracking error and prediction error formed combination misalignment into Row neural network weight updates, and promotes closed loop control process neural network learning performance improvement tracking performance.Due to utilizing nerve Network carry out study can lifting system adaptive ability, reduce robust control method bring conservative, improve control essence Degree；Due to carrying out mind using online data structure forecast error assessment neural network learning performance and in conjunction with system tracking error Through network weight vector complex updates, neural network learning accuracy and rapidity in closed loop control process are improved, improves and is The tracking accuracy of system.
A kind of the technical solution adopted by the present invention to solve the technical problems: hypersonic aircraft reentry stage neural network Hybrid Learning control method, its main feature is that the following steps are included:
(a) hypersonic aircraft reentry stage kinetic model is established:
The kinetic model includes state variable X=[v, ω]TU=M is inputted with controlc, wherein v=[α β σ]TFor appearance State is angularly measured, and α, β, σ respectively indicate the angle of attack, yaw angle and inclination angle；ω=[p q r]TFor attitude angular rate vector, p, q, r Respectively indicate rolling, pitching and yawrate；Mc=[Mx My Mz]TThe control moment of expression system；I indicates inertia matrix；
(b) X=[x is defined1 x2]T, x1=v, x2=ω.Then Attitude control model may be expressed as:
Wherein g1(x1)=R (), f2(x2)=- I-1Ω I ω, g2(x2)=I-1。
(c) posture angle tracking error e is defined1=x1-yd；Wherein yd=[αd βd σd]TThe guidance generated for guidance system refers to It enables.Design virtual controlling amount are as follows:
Wherein k1∈R3×3To control gain matrix,It can further calculateWherein The first derivative and second dervative respectively guidanceed command.
Define attitude angular rate errorDesign control signal McAre as follows:
WhereinFor the estimated value of optimal neural network weight vectors, θ2(x2) it is radial basis function vector；k2∈R3×3For Gain matrix is controlled,
(d) it defines Wherein τd> 0 is integrating range.
Structure forecast error isNeural network Hybrid Learning adaptive lawDesign are as follows:
Wherein λ2∈R3×3To learn rate matrix,kω2∈R3×3For weight factor matrix,
(e) according to obtained control input Mc, back to hypersonic aircraft reentry stage kinetic model (1), (2), tracing control is carried out to attitude angle.
The beneficial effects of the present invention are: hypersonic aircraft reentry stage kinetic description is first towards control by this method The attitude mode of system recycles neural network to carry out learning scene controller to system nondeterministic function, finally using in line number According to structure forecast modeling error, and combination misalignment is formed using tracking error and prediction error and carries out the update of neural network weight, Promote closed loop control process neural network learning performance improvement tracking performance.It is due to carrying out learning to be promoted using neural network The adaptive ability of system reduces robust control method bring conservative, improves control precision；Due to utilizing online data structure It makes prediction error assessment neural network learning performance and combines system tracking error, it is compound more to carry out neural network weight vectors Newly, neural network learning accuracy and rapidity in closed loop control process are improved, the tracking accuracy of system is improved.
It elaborates with reference to the accompanying drawings and detailed description to the present invention.
Detailed description of the invention
Fig. 1 is the flow chart of hypersonic aircraft reentry stage neural network Hybrid Learning control method of the present invention.
Specific embodiment
Referring to Fig.1.Hypersonic aircraft reentry stage neural network Hybrid Learning control method specific steps of the present invention are such as Under:
(a) hypersonic aircraft reentry stage kinetic model is established:
The kinetic model includes state variable X=[v, ω]TU=M is inputted with controlc, wherein v=[α β σ]TFor appearance State is angularly measured, and α, β, σ respectively indicate the angle of attack, yaw angle and inclination angle；ω=[p q r]TFor attitude angular rate vector, p, q, r Respectively indicate rolling, pitching and yawrate；Mc=[Mx My Mz]TThe control moment of expression system；
(b) X=[x is defined1 x2]T, x1=v, x2=ω.Then Attitude control model may be expressed as:
Wherein g1(x1)=R (), f2(x2)=- I-1Ω I ω, g2(x2)=I-1。
(c) posture angle tracking error e is defined1=x1-yd；Wherein yd=[αd βd σd]TThe guidance generated for guidance system refers to It enables；Design virtual controlling amount are as follows:
WhereinIt can further calculateWhereinRespectively make Lead the first derivative and second dervative of instruction.
Define attitude angular rate errorDesign control signal McAre as follows:
WhereinFor the estimated value of optimal neural network weight, θ2(x2) it is RBF basis function vector；
(d) it defines Wherein τd=0.05s.
Structure forecast error isDesign neural network Hybrid Learning adaptive lawAre as follows:
Wherein
(e) according to obtained control input Mc, back to hypersonic aircraft reentry stage kinetic model (1), (2), tracing control is carried out to attitude angle.
Unspecified part of the present invention belongs to field technical staff's common knowledge.

Claims (1)

1. a kind of hypersonic aircraft reentry stage neural network Hybrid Learning control method, it is characterised in that including following step It is rapid:
(a) hypersonic aircraft reentry stage kinetic model is established:
The kinetic model includes state variable X=[v, ω]TU=M is inputted with controlc, wherein v=[α β σ]TFor attitude angle Vector, α, β, σ respectively indicate the angle of attack, yaw angle and inclination angle；ω=[p q r]TFor attitude angular rate vector, p, q, r difference Indicate rolling, pitching and yawrate；Mc=[Mx My Mz]TThe control moment of expression system；I indicates inertia matrix；
(b) X=[x is defined1 x2]T, x1=v, x2=ω；Then Attitude control model may be expressed as:
Wherein g1(x1)=R (), f2(x2)=- I-1Ω I ω, g2(x2)=I-1；
(c) posture angle tracking error e is defined1=x1-yd；Wherein yd=[αd βd σd]TIt is guidanceed command for what guidance system generated； Design virtual controlling amount are as follows:
Wherein k1∈R3×3To control gain matrix,It can further calculateWherein The first derivative and second dervative respectively guidanceed command；
Define attitude angular rate errorDesign control signal McAre as follows:
WhereinFor the estimated value of optimal neural network weight vectors, θ2(x2) it is radial basis function vector；k2∈R3×3For control Gain matrix,
(d) it definesIts Middle τd> 0 is integrating range；
Structure forecast error isNeural network Hybrid Learning adaptive lawDesign are as follows:
Wherein λ2∈R3×3To learn rate matrix,kω2∈R3×3For weight factor matrix,
(e) according to obtained control input Mc, back to the kinetic model (1) of hypersonic aircraft reentry stage, (2), to appearance State angle carries out tracing control.
CN201710789276.6A 2017-09-05 2017-09-05 Hypersonic aircraft reentry stage neural network Hybrid Learning control method Active CN107450323B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710789276.6A CN107450323B (en) 2017-09-05 2017-09-05 Hypersonic aircraft reentry stage neural network Hybrid Learning control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710789276.6A CN107450323B (en) 2017-09-05 2017-09-05 Hypersonic aircraft reentry stage neural network Hybrid Learning control method

Publications (2)

Publication Number Publication Date
CN107450323A CN107450323A (en) 2017-12-08
CN107450323B true CN107450323B (en) 2019-10-18

Family

ID=60495139

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710789276.6A Active CN107450323B (en) 2017-09-05 2017-09-05 Hypersonic aircraft reentry stage neural network Hybrid Learning control method

Country Status (1)

Country Link
CN (1) CN107450323B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109164708B (en) * 2018-10-31 2020-08-04 南京航空航天大学 Neural network self-adaptive fault-tolerant control method for hypersonic aircraft
CN110413000B (en) * 2019-05-28 2020-10-13 北京航空航天大学 Hypersonic aircraft reentry prediction correction fault-tolerant guidance method based on deep learning
CN111351488B (en) * 2020-03-03 2022-04-19 南京航空航天大学 Intelligent trajectory reconstruction reentry guidance method for aircraft
CN112327626B (en) * 2020-11-14 2022-06-21 西北工业大学 Aircraft channel coupling coordination control method based on data analysis
CN112327627B (en) * 2020-11-14 2022-06-21 西北工业大学 Nonlinear switching system self-adaptive sliding mode control method based on composite learning
CN114859712B (en) * 2022-04-17 2023-08-01 西北工业大学 Aircraft guidance control integrated method oriented to throttle constraint

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2423377A (en) * 2002-12-09 2006-08-23 Georgia Tech Res Inst Adaptive output feedback apparatuses and methods capable of controlling a non-minimum phase system.
US8285659B1 (en) * 2009-08-18 2012-10-09 The United States of America as represented by the Administrator of the National Aeronautics & Space Administration (NASA) Aircraft system modeling error and control error
CN102880055A (en) * 2012-09-29 2013-01-16 西北工业大学 Method for controlling neural network of hypersonic aerocraft on basis of prediction model
CN102880052A (en) * 2012-09-29 2013-01-16 西北工业大学 Time scale function decomposition based hypersonic aircraft actuator saturation control method
CN102929141A (en) * 2012-10-10 2013-02-13 西北工业大学 Aircraft time lag varying model approximation and controller designing method
CN103365296A (en) * 2013-06-29 2013-10-23 天津大学 Nonlinear output feedback flight control method for quad-rotor unmanned aerial vehicle
CN104022742A (en) * 2014-06-19 2014-09-03 金陵科技学院 Aircraft posture robust inversion fault tolerant control method based on neural network observer
CN104765272A (en) * 2014-03-05 2015-07-08 北京航空航天大学 Four-rotor aircraft control method based on PID neural network (PIDNN) control
US9146557B1 (en) * 2014-04-23 2015-09-29 King Fahd University Of Petroleum And Minerals Adaptive control method for unmanned vehicle with slung load

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050137724A1 (en) * 2003-10-10 2005-06-23 Georgia Tech Research Corporation Adaptive observer and related method

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2423377A (en) * 2002-12-09 2006-08-23 Georgia Tech Res Inst Adaptive output feedback apparatuses and methods capable of controlling a non-minimum phase system.
US8285659B1 (en) * 2009-08-18 2012-10-09 The United States of America as represented by the Administrator of the National Aeronautics & Space Administration (NASA) Aircraft system modeling error and control error
CN102880055A (en) * 2012-09-29 2013-01-16 西北工业大学 Method for controlling neural network of hypersonic aerocraft on basis of prediction model
CN102880052A (en) * 2012-09-29 2013-01-16 西北工业大学 Time scale function decomposition based hypersonic aircraft actuator saturation control method
CN102929141A (en) * 2012-10-10 2013-02-13 西北工业大学 Aircraft time lag varying model approximation and controller designing method
CN103365296A (en) * 2013-06-29 2013-10-23 天津大学 Nonlinear output feedback flight control method for quad-rotor unmanned aerial vehicle
CN104765272A (en) * 2014-03-05 2015-07-08 北京航空航天大学 Four-rotor aircraft control method based on PID neural network (PIDNN) control
US9146557B1 (en) * 2014-04-23 2015-09-29 King Fahd University Of Petroleum And Minerals Adaptive control method for unmanned vehicle with slung load
CN104022742A (en) * 2014-06-19 2014-09-03 金陵科技学院 Aircraft posture robust inversion fault tolerant control method based on neural network observer

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
《Neural Control for Longitudinal Dynamics of Hypersonic Aircraft》;Bin Xu;《2013 International Conference on Unmanned Aircraft Systems (ICUAS)》;20130531;全文 *
《RBF Neural Network based Adaptive Sliding Mode Control for Hypersonic Flight Vehicles》;Jianmin Wang;《Proceedings of 2016 IEEE Chinese Guidance, Navigation and Control Conference》;20160815;全文 *
《基于RBF神经网络自适应PID四旋翼飞行器控制》;李砚浓;《控制工程》;20160331;第23卷(第3期);全文 *
《基于混合神经网络的鲁棒自适应飞行控制器的设计》;王丽;《电光与控制》;20161130;第23卷(第11期);全文 *

Also Published As

Publication number Publication date
CN107450323A (en) 2017-12-08

Similar Documents

Publication Publication Date Title
CN107450323B (en) Hypersonic aircraft reentry stage neural network Hybrid Learning control method
CN108803649B (en) A kind of VTOL Reusable Launch Vehicles active disturbance rejection sliding-mode control
CN106527137B (en) Quadrotor drone fault tolerant control method based on observer
Wu et al. Modified adaptive neural dynamic surface control for morphing aircraft with input and output constraints
Chen et al. A novel nonlinear resilient control for a quadrotor UAV via backstepping control and nonlinear disturbance observer
CN107608367B (en) Multivariable interference compensation quadrotor unmanned aerial vehicle trajectory and attitude cooperative control method
CN103838145B (en) VTOL aircraft Robust Fault-Tolerant Control Systems based on cascade observer and method
CN104155990B (en) Consider the hypersonic aircraft pitch channel attitude control method of angle of attack constraint
CN105607473B (en) The attitude error Fast Convergent self-adaptation control method of small-sized depopulated helicopter
CN104281155B (en) Three-dimensional flight path tracking method for unmanned airship
Xu et al. Adaptive prescribed performance terminal sliding mode attitude control for quadrotor under input saturation
CN104950898A (en) Reentry vehicle full-order non-singular terminal sliding mode posture control method
CN107526296B (en) Hypersonic aircraft neural network lea rning control method based on prediction modeling
CN113361013B (en) Spacecraft attitude robust control method based on time synchronization stability
CN101708780A (en) Method for controlling rigid spacecraft for target attitude tracking
CN108594837A (en) Model-free quadrotor drone contrail tracker and method based on PD-SMC and RISE
CN113377121B (en) Aircraft intelligent disturbance rejection control method based on deep reinforcement learning
CN105242683B (en) A kind of dirigible neutral net terminal sliding mode flight tracking control method
CN107065564A (en) A kind of neutral buoyancy robot pose and method for controlling trajectory based on active disturbance rejection
CN108803648A (en) Unmanned vehicle robust attitude control method, device and electronic equipment
Zhu et al. Tracking control for angular-rate-sensorless vertical take-off and landing aircraft in the presence of angular-position measurement delay
CN109164708A (en) A kind of hypersonic aircraft neural network adaptive fusion method
CN105116914A (en) Stratospheric-airship-analytic-model-based prediction path tracking control method
CN107479382B (en) Hypersonic aircraft neural network control method based on online data study
CN108646554A (en) A kind of quick anti-interference longitudinal guidance method of the aircraft based on specified performance

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant